1 / 1

Do Integer Ratings Hinder Recommendation Systems?

Do Integer Ratings Hinder Recommendation Systems? By: Shi Hui Lim, Jimmy Bobowski , Amanda Lambert. Introduction Problem: Do integer ratings hinder recommendation systems? Use Matrix Completion algorithm to complete sparse matrices that come from the original Netflix training dataset

errol
Download Presentation

Do Integer Ratings Hinder Recommendation Systems?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Do Integer Ratings Hinder Recommendation Systems? By: Shi Hui Lim, Jimmy Bobowski, Amanda Lambert • Introduction • Problem: Do integer ratings hinder recommendation systems? • Use Matrix Completion algorithm to complete sparse matrices that come from the original Netflix training dataset • Compare matrices using RMSE (Root Mean Squared Error) • Determine how much error is inherent in Netflix’s rounded rating recommendation system • Methodology • Create five initial matrices: • M0 = Create initial matrix from Netflix • training dataset • M = Add ‘noise’ to M0 (i.e. a rating of • ‘3’ becomes ‘3.212’) • Ms = Randomly remove entries from • M • Mr = Round the entries in Ms • MR = Round the entries in M • Apply Matrix completion algorithm on Ms and Mr to obtain two complete matrices (Ms~ and Mr~) • Compare Ms~ and Mr~ to M and MR using RMSE Plot Theory Matrix Completion: works well for sparse matrices with a reasonable (~10%) amount of entries and can recreate a full matrix with high probability of success RMSE: Hypothesis The higher the rank in a matrix, the less successful the Matrix Completion algorithm will be When there are less entries in a matrix, the Matrix Completion algorithm will have more error The Matrix Completion algorithm will be more successful in recreating matrices with unrounded numbers Conclusion We were able to show that there is a certain amount of error in Netflix’s recommendation We ran into problems with the Matrix Completion algorithm that we used and were unable to show some of our other hypotheses Rounded user ratings will always have more error than unrounded data because in a recommendation system sdfaasdfasdfasdfasdfasdfadfsdfsdfasdfsdfsdfasdfsdfsdfsdfsdfsdfsdfsdf 2s m x n Mentors: Brent Castle, Huijun Wang I399 – Research Methods

More Related